Академический Документы
Профессиональный Документы
Культура Документы
|
Revised: 7 February 2020
| Accepted: 13 February 2020
DOI: 10.1111/epi.16469
1
Department of Anesthesiology and Critical
Care Medicine, Children’s Hospital of
Abstract
Philadelphia, Philadelphia, Pennsylvania Objective: To compare the safety and efficacy of phenobarbital and levetiracetam in
2
Departments of Neurology and Pediatrics, a cohort of neonates with seizures following cardiac surgery.
Children’s Hospital of Philadelphia,
Methods: We performed a retrospective single-center study of consecutive
University of Pennsylvania, Philadelphia,
Pennsylvania neonates with electrographically confirmed seizures managed with antiseizure
3
Division of Cardiothoracic Surgery, medication after cardiac surgery from June 15, 2012 to December 31, 2018. We
Children’s Hospital of Philadelphia, compared the safety and efficacy of phenobarbital and levetiracetam as first-line
Philadelphia, Pennsylvania
4
therapy.
Department of Biostatistics, Epidemiology,
and Informatics, University of Pennsylvania Results: First-line therapy was phenobarbital in 31 neonates and levetiracetam in
School of Medicine, Philadelphia, 22 neonates. Phenobarbital was associated with more adverse events (P = .006).
Pennsylvania
Eight neonates (14%) experienced an adverse event related to phenobarbital use,
Correspondence including seven with hypotension and one with respiratory depression. No ad-
Shavonne L. Massey, Division of verse events were reported with levetiracetam use. The cessation of electrographic
Neurology, Children's Hospital of
seizures was similar in both groups, including 18 neonates (58%) with seizure
Philadelphia, Philadelphia, PA 19104.
Email: masseysl@email.chop.edu cessation after phenobarbital and 12 neonates (55%) with seizure cessation after
levetiracetam (P = 1.0). The combined cessation rates of phenobarbital and lev-
Funding information
National Institute of Neurological Disorders
etiracetam when used as first- or second-line therapy were 58% and 47%, respec-
and Stroke, Grant/Award Number: NS- tively (P = .47).
072338 and NS-096058; Wolfson, Grant/ Significance: Phenobarbital was associated with more adverse events than lev-
Award Number: NS-072338
etiracetam, and the two drugs were equally but incompletely effective in treat-
ing electrographically confirmed seizures in neonates following cardiac surgery.
Given its more acceptable safety profile and potential noninferiority, leveti-
racetam may be a reasonable option for first-line therapy for treatment of sei-
zures in this population. Further prospective studies are needed to confirm these
results.
KEYWORDS
cardiology, congenital, critical care, heart diseases, infant
Epilepsia. 2020;00:1–9. |
wileyonlinelibrary.com/journal/epi 1
|
2 THIBAULT et al.
1 | IN TRO D U C T ION
Key Points
Neonates with congenital heart disease (CHD) undergoing
cardiac surgery with cardiopulmonary bypass (CPB) are at • Phenobarbital was associated with more adverse
risk for seizures, with a reported incidence ranging between events than levetiracetam
5% and 20%.1–7 The occurrence of postoperative seizures is • Phenobarbital and levetiracetam were equally but
associated with in-hospital mortality and worse neurodevel- incompletely effective in treating neonatal sei-
opmental outcomes.1,8–10 Timely and effective management zures following cardiac surgery
of postoperative seizures may reduce seizure-induced sec- • Levetiracetam may be a reasonable option for
ondary brain injury, yet the optimal antiseizure medication first-line treatment for seizures in this population
(ASM) strategy to treat acute symptomatic seizures follow- • Further prospective studies are needed to confirm
ing CHD surgery in newborns remains unknown. these results
Phenobarbital (PHB) has been used for more than a cen-
tury and is the traditional first-line ASM for neonatal sei-
zures.11 It is metabolized by the liver via cytochrome P450
and primarily excreted by the kidneys.12 PHB pharmacoki- recommendations,26 all patients were tracked in a clinical da-
netics is notable for maturation-related changes in clearance tabase as part of a previously published quality improvement
in the first months of life, auto-inducible clearance with use, project.1 Only neonates with EEG-confirmed postoperative
and numerous drug interactions.12,13 Prior studies in neonates, seizures were included in this study. Per our protocol, cEEG
including randomized controlled trials, have reported PHB was continued for at least 24 hours after electrographic sei-
efficacy of 40%-50% for neonatal seizures, although success zure cessation. This study was approved by the Children's
rates as high as 72% have been reported.14–17 Levetiracetam Hospital of Philadelphia Institutional Review Board.
(LEV) is a newer ASM increasingly used to treat neonatal
seizures.18 A small fraction of the administered dose is me-
tabolized via nonhepatic pathways, whereas two-thirds is ex- 2.2 | Characterization of seizures and
creted unchanged via the kidneys. LEV pharmacokinetics is response to treatment
notable for a linear clearance influenced by renal function
and subject to maturation-related changes in the first weeks of For the purpose of this study, all EEG tracings were rein-
life, few drug interactions, and a wide therapeutic index.19,20 terpreted by a single electroencephalographer (S.L.M.).
Its efficacy in controlling seizures in neonates has been re- Electrographic seizures were defined as sudden and abnor-
ported to be between 35% and 100%. However, the data are mal EEG events with a repetitive and evolving pattern with a
derived from observational studies with small sample sizes, minimum 2-µV peak-to-peak voltage and duration of at least
and efficacy was not consistently assessed with continuous 10 seconds.27 Seizures were classified as EEG-only (subclini-
electroencephalographic monitoring (cEEG), which is the cal, nonconvulsive) if no clinical signs were observed by the
gold standard for neonatal seizure detection.17,21–25 To date, bedside provider or on video review, or electroclinical if a
studies have not established an optimal seizure management clinical manifestation was associated with an ictal discharge.
strategy in neonates with CHD. Thus, we aimed to describe Clinical-only seizures (ie, seizures lacking an electrographic
the safety and efficacy of PHB and LEV in treating postop- correlate) were not included. Seizures were further classi-
erative electrographically confirmed acute symptomatic sei- fied according to their frequency as: (1) rare (<5 seizures),
zures in neonates with CHD. (2) occasional (5-10 seizures), (3) frequent (>10 seizures),
or (4) status epilepticus (summed duration of electrographic
seizures comprises >50% of a 1-hour epoch). Our institu-
2 | M AT E R IA L S A N D ME T HODS tion had no standardized practice for seizure management in
neonates following cardiac surgery, and ASM choices were
2.1 | Patient population made by the clinical team. Favorable response to treatment
was defined as (1) complete cessation of electrographic sei-
This was a single-center retrospective study of consecu- zures following ASM administration and (2) no administra-
tive neonates (≤30 days of age, corrected gestational age ≤ tion of further ASMs for seizure control. If an ASM was
44 weeks) with electrographically confirmed seizures fol- successful in terminating seizures but was discontinued for
lowing cardiac surgery from June 15, 2012 to December safety concerns, then it was defined as effective. If ≥2 ASMs
31, 2018. Following implementation of routine 48-hour were administered before seizure cessation, then the last ad-
cEEG in all neonates following cardiac surgery with CPB ministered ASM was defined as effective, unless otherwise
as per the American Clinical Neurophysiology Society specified in the medical record.
THIBAULT et al.
|
3
2.3 | Data collection and the clinical team decided treatment was not war-
ranted; these neonates were excluded from further analysis.
We collected data including (1) patient demographics (ges- Among the 56 neonates who received ≥1 ASM, the median
tational age, birth weight, age at surgery), (2) surgery-re- gestational age, birth weight, and postnatal age at surgery
lated data (primary cardiac anomaly, procedure performed, were 38 weeks (37-39), 3.2 kg (2.7-3.5), and 5 days (3-8),
CPB, deep hypothermic circulatory arrest and aortic respectively. Nineteen neonates (33.9%) were premature
cross-clamp times, extracorporeal membrane oxygena- (<37 weeks of gestational age). The underlying cardiac
tion [ECMO] support while on cEEG), (3) seizure type anatomy was a single-ventricle defect in 55.4% of neonates
(EEG-only, electroclinical) and severity (status epilepticus (31 of 56). The four most common procedures performed
or not), and (4) ASM data (sequence of ASM administra- were the Norwood (20 of 56; 35.7%), arterial switch (six of
tion, timing, dosing, duration of therapy, relationship to 56; 10.7%), systemic to pulmonary artery shunt (five of 56;
electrographic seizure cessation, plasmatic concentrations 8.9%), and repair of truncus arteriosus (five of 56; 8.9%).
obtained for therapeutic drug monitoring as per standard Fourteen neonates (25.0%) were on ECMO, with 43% (six
of care). Loading doses were defined as any ASM boluses of 14) cannulated in the operating room and 57% (eight of
given with the intent of quickly reaching a therapeutic 14) cannulated in the cardiac intensive care unit (CICU).
steady state and ending seizures within the initial 72 hours Seven neonates (seven of 14; 50.0%) were cannulated to
following seizure onset. The total loading amount was de- ECMO following a cardiac arrest, including two neonates
fined as the sum of all loading doses received over 72 hours. in the operating room (two of six; 33.3%) and five neonates
Maintenance doses were defined as scheduled doses aimed in the CICU (five of eight; 62.5%). The remaining patients
to maintain a steady state. Peak PHB plasma concentra- were electively cannulated. EEG-only seizures were pre-
tions were defined as concentrations obtained 1-2 hours sent in 98.2% of neonates with seizures (55 of 56), and
following a loading dose. Because of their antiseizure electroclinical seizures were present in 14.3% of neonates
properties, data were also collected regarding concomitant with seizures (eight of 56). Status epilepticus occurred in
administration of midazolam and ketamine for sedation or 23.2% of neonates with seizures (13 of 56). Sixteen (29%)
analgesia. Adverse events were defined as any untoward neonates died during their hospitalization. None of the
event attributed to an ASM by the clinical team and docu- deaths was considered to be related to ASM.
mented in the medical record. Collected safety data were
limited to the presence or lack of a specific adverse event
as a dichotomous variable. Additionally, deaths occurring 3.2 | ASM administration as
during patient hospitalization were recorded, regardless of first-line therapy
suspected relationship to ASM use.
Among the neonates who were treated with ≥1 ASM,
55.3% (31 of 56) received PHB, 39.2% (22 of 56) received
2.4 | Statistical analysis LEV, 3.5% (two of 56) received lorazepam, and 1.8% (one
of 56) received ketamine as first-line therapy. Table 1 pro-
We performed statistical analysis using the software Stata vides a comparison of the demographics and clinical char-
(v14.2). We reported summary statistics as medians and in- acteristics between neonates who received PHB or LEV as
terquartile ranges for continuous data and proportions for cat- first-line therapy. The two groups were similar in baseline
egorical data. Comparative statistics for neonates receiving characteristics.
PHB versus LEV as first-line ASM therapy were performed
with Fisher exact test for categorical variables and Wilcoxon
rank-sum test for skewed continuous variables. 3.3 | ASM adverse events
3.4 | ASM efficacy P = .35). Of the 13 patients who did not respond to PHB as
first-line therapy, 10 (76.9%) received LEV as second-line
The control of electrographic seizures following first-line ASM therapy with seizure cessation in three (30.0%), one (7.7%) re-
administration was similar in both groups; 18 neonates (58.1%) ceived lorazepam with seizure cessation in none (0%), and one
responded to PHB and 12 neonates (54.5%) responded to LEV (7.7%) received phenytoin with seizures cessation in none (0%).
(P = 1.0), as shown in Figure 1. In neonates who responded to One patient died and did not receive a second-line therapy after
first-line ASM therapy, there was no difference in the time from not responding to PHB as first-line therapy. Two neonates who
first ASM dose administration to seizure cessation between maintained control of seizures with PHB were subsequently
PHB and LEV (5.2 hours [0.7-11.6]) vs (1.5 hours [0.4-7.4], transitioned to LEV by the clinical team for long-term use given
THIBAULT et al.
|
5
F I G U R E 1 Flowchart of neonates
receiving phenobarbital or levetiracetam as
first- and second-line therapy and outcomes
(N = 53)
its presumed better safety profile. The combined efficacy of time from seizure onset to first LEV dose administration be-
PHB when used as first- or second-line therapy following LEV tween nonresponders and responders (3.0 hours vs 2.2 hours,
was 57.9%. Of the 10 patients who did not respond to initial P = .25). There was no difference in the LEV initial loading
LEV therapy, seven (70.0%) received PHB as second-line ther- dose amount, number of loading doses, total loading amount,
apy with seizure cessation in four (57.1%), one (10%) received or time to administer all loading doses administered for first-
phenytoin with seizure cessation in none (0%), and two (20%) line LEV nonresponders and responders (initial loading dose:
received lorazepam with seizure cessation in none (0%). The 32.5 mg/kg [20-50] vs 30 mg/kg [20-30], P = .54; number of
combined efficacy of LEV when used as first- or second-line doses: 1.5 [1-2] vs 1 [1-1.5], P = .55; total loading amount:
therapy following PHB was 46.9%. LEV and PHB combined 50 mg/kg [40-60] vs 30 mg/kg [20-50.5], P = .10; time to ad-
(first- and second-line) efficacy was not different (P = .47). minister all loading doses: 0.33 hours [0-8.23] vs 0 hours [0-
LEV and PHB were also administered as third-line therapy 2.75], P = .42). Neonates on ECMO received a higher PHB
in four and three neonates, respectively. Excluding PHB and total loading amount than neonates not supported by ECMO
LEV, other ASMs administered included lorazepam in six neo- (50 mg/kg [30-55] vs 25 mg/kg [20-40], P = .03). Similarly,
nates (10.7%), midazolam infusion in seven neonates (12.5%), neonates with status epilepticus received a higher PHB total
midazolam boluses in three neonates (5.4%), fosphenytoin in loading amount than neonates with a lower seizure burden
four neonates (7.1%), phenytoin in two neonates (3.6%), and (50 mg/kg [30-55] vs 25 mg/kg [20-46.4], P = .05). Neonates
ketamine as a combination of infusion and boluses in two neo- with or without ECMO, and neonates with or without status
nates (3.6%). Figure 1 summarizes outcomes for neonates who epilepticus, did not receive different LEV total loading amount
received either PHB or LEV as first-line ASM. Figure S1 sum- (ECMO: 44.6 mg/kg [20-69.1] and 45 mg/kg [30-51], P = .09;
marizes outcomes for each ASM administered for the entire status epilepticus: 50 mg/kg [40-60] vs 40 mg/kg [20-51],
cohort, including non-PHB/LEV first-line ASMs. P = .31). Of the 31 neonates who received PHB as first-line
therapy, 20 neonates (65%) had peak PHB levels drawn per
standard of care 1-2 hours following a loading dose. There was
3.5 | ASM dosing no difference between maximum peak PHB concentrations in
neonates with or without favorable response to PHB first-line
Table 2 summarizes the dosing, treatment timing, and peak therapy (29.4 mg/L [22.6-45.5] vs 40.1 mg/L [24.6-46.2],
concentrations for patients who received PHB or LEV as P = .57), with or without ECMO support (24.6 mg/L [20.3-
first-line treatment. The time from seizure onset to first PHB 43.1] vs 40.4 mg/L [27.7-45.7], P = .17), or with or without
dose administration was shorter in PHB nonresponders than status epilepticus (44.7 mg/L [24.6-47.1] vs 33.2 mg/L [22.6-
responders (1.9 hours vs 2.9 hours, P = .005). The initial load- 44.5], P = .19). LEV therapeutic levels were not routinely col-
ing dose was not different between PHB nonresponders and lected as standard clinical care.
responders (10 mg/kg [10-20] vs 20 mg/kg [10-20], P = .2).
First-line PHB nonresponders received significantly more
PHB boluses as loading dose than responders (4 [3-6] vs 2 [1- 3.6 | Concomitant medications during
2], P = .00). As a result, first-line PHB nonresponders received loading dose administrations
a significantly higher total loading amount than responders
(47.8 mg/kg [30-50] vs 20 mg/kg [19.2-25], P = .005) over a Some neonates were administered medications with antisei-
longer period of time (12.63 hours [11.00-30.22] vs 0.35 hours zure properties for nonseizure purposes during the initial 72-
[0-5.75], P = .003). There was no difference between the hour period following seizure onset. Midazolam infusions
6
| THIBAULT et al.
were administered for sedation in 7% of neonates (four of 56) response to first-line ASM administration. The pres-
at a maximum dose of 0.03 mg/kg/h (0.02-0.04). Ketamine ence of a single-ventricle defect was also associated
infusions were administered for sedation and analgesia in with decreased likelihood of response to first-line
14.3% (eight of 56) at a maximum dose of 0.25 mg/kg/h ASM therapy (P = .01). There was no difference be-
(0.15-0.30). There was no difference in concomitant mida- tween time from seizure onset to first ASM administra-
zolam infusions between neonates who received PHB or tion between responders and nonresponders (P = .08).
LEV as first-line therapy (P = .64), whereas more neonates There was no difference between the time from seizure
who received LEV as first-line therapy received concomitant onset to first ASM administration between PHB and
low-dose ketamine infusion (six of 22 [27.3%] vs two of 31 LEV (2.32 hours [1.50-4.12] vs 2.27 hours [1.87-5.18],
[6.5%], P = .05). Concomitant ketamine or midazolam infu- P = .59). Cessation of seizures following administra-
sions were not associated with more favorable response to tion of a first-line ASM was associated with survival to
PHB (midazolam: P = .50, ketamine: P = 1.0) or LEV (mi- discharge (86% in responders vs 52% in nonresponders;
dazolam: P = .46, ketamine: P = .65) as first-line therapies. P = .03).
One (1.9%) patient received intermittent midazolam boluses PHB nonresponders were more likely to be on ECMO and
of 0.03 mg/kg/dose for sedation (total of 0.27 mg/kg admin- to have status epilepticus than responders (ECMO: eight of
istered over 43.05 hours). Boluses of ketamine and intermit- 13 [61.5%] vs two of 18 [11.1%], P = .006; status epilepticus:
tent doses of lorazepam were not administered for sedation nine of 13 [69%] vs one of 18 [5.6%], P = .00). The propor-
and analgesia. tion of single-ventricle defects did not differ between PHB
nonresponders and responders (nine of 13 [69%] vs seven of
18 [39%], P = .15). There was no difference in the propor-
3.7 | Associated clinical features tions of patients on ECMO or with status epilepticus between
the LEV nonresponders and responders (ECMO: two of 10
Lower seizure burden (P = .001) and absence of ECMO [20%] vs 0 of 12 [0%], P = .20; status epilepticus: two of
support (P = .002) were associated with favorable 10 [20%] vs one of 12 [8%], P = .57). LEV nonresponders
THIBAULT et al.
|
7
were more likely to have a single-ventricle defect (eight of 10 LEV for less complicated clinical situations, and this may
[80%] vs four of 12 [33.3%], P = .04). have negatively impacted PHB success rate. Moreover, a
higher underlying illness burden may have contributed to the
increased incidence of adverse effect associated with PHB.
4 | D IS C U SSION Although few neonates concomitantly received a ketamine
infusion, the proportion was higher among neonates who re-
Electrographic seizures occurred in 8.4% of neonates fol- ceived LEV as first-line ASM. Ketamine-LEV combination,
lowing cardiac surgery with CPB, consistent with published either through additive or synergistic effects, may have re-
literature.1,5,28 PHB and LEV were the most commonly ad- sulted in greater antiseizure efficacy, thereby increasing the
ministered first-line ASMs. Seizures ceased after first-line LEV success rate.
ASM therapy in about half of patients, consistent with previ- Although our data did not indicate a better response rate
ous studies in neonatal populations.14,15,17 Efficacy was not in patients with shorter duration between seizure onset and
associated with the choice of first-line ASM (PHB or LEV), treatment, faster ASM administration has been associated
but it was associated with seizure burden and underlying dis- with better outcomes in previous studies.40,41 Paradoxically,
ease severity. Factors associated with less favorable response PHB nonresponders had a shorter duration between seizures
to first-line ASM therapy included ECMO exposure and sin- and ASM administration in our study. We believe the benefi-
gle-ventricle defects, which are both independently associ- cial effect of prompt treatment may have been masked in the
ated with an increased risk of seizures.2,29,30 current study because (1) PHB nonresponders were perceived
PHB safety profile may limit its use in neonates following by the clinical team as more clinically unstable; and (2) PHB
cardiac surgery. Short-term adverse events, including hypo- nonresponders had more refractory seizures, as shown by the
tension and respiratory depression, are concerning in patients greater proportions of status epilepticus in this cohort and a
for whom hemodynamic instability frequently occurs concur- trend toward more ECMO exposure. Better anticipation of
rently with seizures. In our cohort, 16.3% of neonates who seizures in these high-risk patients may explain faster access
received PHB had hypotension attributed to its use, although to medication, and the harder-to-treat nature of the seizures
this may be an underestimation due to the retrospective nature may explain the lower treatment response.
of our study. To our knowledge, the incidence of hypotension Our study has several limitations. First, the size and ret-
associated with PHB administration in neonates following car- rospective nature of our study may have prevented iden-
diac surgery has not been previously reported. However, up to tification of specific subgroups who may have a better
53% of neonates with hypoxic-ischemic encephalopathy were response to a specific ASM (eg, neonates on ECMO). The
prospectively found to have clinically significant hypotension retrospective nature of our study also prevented us from
requiring inotropes after receiving PHB.31 Furthermore, neg- characterizing seizure burden with precision (ie, in min-
ative long-term consequences associated with PHB use have utes per hour). As this meaningful information was not col-
been described, including animal data indicating PHB-induced lected in our database, we had to rely on seizure frequency,
synaptic changes and neuronal apoptosis in the neonatal which is less informative. Second, the small sample size
brain.32,33 LEV has not been associated with serious adverse may impact the ability to detect a small difference in effi-
events in clinical studies.22,23,34–36 Moreover, it has not been cacy between PHB and LEV. We continue to collect stan-
associated with cellular apoptosis in animal models.37,38 dardized data on our cohort, which will allow continuous
LEV could be considered as a first-line ASM given our assessment of efficacy in a larger cohort. Third, the optimal
data indicating equal efficacy with PHB and a more favor- PHB and LEV doses are not well defined in postoperative
able safety profile. Despite a lack of large prospective tri- neonates, and suboptimal ASM dosing in some neonates
als, multiple small studies have demonstrated LEV efficacy may have led to reduced efficacy. PHB nonresponders re-
in neonates.17,21–25,35 Moreover, McHugh et al39 performed ceived significantly more loading doses and a higher PHB
a systematic review to compare the effectiveness of PHB to total loading amount. This may reflect practice, in which
LEV in a heterogenous cohort of neonates with seizures and clinicians administer additional PHB loading doses to
did not find any significant difference between the ASMs. nonresponders before initiating another ASM. However,
However, some factors need to be considered when inter- despite nonresponders receiving a higher total PHB load,
preting our results. There was a nonsignificant trend toward there were no significant differences in peak PHB levels
administering PHB first in status epilepticus. Similarly, between neonates with or without a favorable response
neonates who were more medically unstable seemed more to therapy, with or without ECMO exposure, and with or
likely to receive PHB first, as suggested by a trend toward a without status epilepticus. This study was not designed to
higher proportion of ECMO among neonates who received characterize pharmacokinetic variability, and this observa-
PHB initially. These differences may reflect a clinician bias tion is limited by very sparse sampling collected at differ-
toward using PHB for presumed difficult to treat seizures and ent time points. However, it suggests some interindividual
|
8 THIBAULT et al.